Hindawi

Mobile Information Systems

Volume 2020, Article ID 8875301, 15 pages
https://doi.org/10.1155/2020/8875301

Research Article
A Smart Parking System Based on Mini PC Platform and Mobile
Application for Parking Space Detection

Vladimir Sobeslav and Josef Horalek

Department of Information Technologies, Faculty of Informatics and Management, University of Hradec Kralove,
Rokitanskeho 62, Hradec Kralove 500 01, Czech Republic

Correspondence should be addressed to Vladimir Sobeslav; [email protected]

Received 24 June 2020; Revised 25 September 2020; Accepted 9 October 2020; Published 26 October 2020

Academic Editor: Peter Brida

Copyright © 2020 Vladimir Sobeslav and Josef Horalek. This is an open access article distributed under the Creative Commons
Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is
properly cited.

Car parking is a major problem in urban areas in developed and also in developing countries. The growing number of vehicles
creates a problem with parking spaces mainly in the city center and the surrounding streets. The local authorities have to react with
regulations, and the current situation is unpleasant for many citizens. Therefore, the aim of this article is to propose a complex
outdoor smart parking lot system based on the mini PC platform with the pilot implementation, which would provide a solution
for the aforementioned problem. Current outdoor car park management is dependent on human personnel keeping track of the
available parking lots or a sensor-based system that monitors the availability of each car. The proposed solution utilizes a modern
IoT approach and technologies such as mini PC platform, sensors, and IQRF. When compared to a specialized and expensive
system, it is a solution that is cost-effective and has the potential in its expansion and integration with other IoT services.

1. Introduction with its pilot implementation, which would provide a so-

lution for the aforementioned problem. The solution aims to
The number of vehicles is constantly increasing, not only in use the newest principles in the IoTarea, mesh networks, and
the Czech Republic but also in other countries. According to tools offered by the Android OS as for the whole solution to
the data from Central Auto-moto Club of the Czech Re- be affordable and quickly deployable.
public, more than 5.5 million cars are registered here, which The proposed solution contains a complete design and
means that their number has increased 2.4 times since 1989. realization based on the mini PC platform, IQRF technology,
With the increasing number of the vehicles, the problems including use of the DPA protocol. As a gateway, UpBoard
with the parking also arise. First and foremost, these with DK-EVAL-04A communication module has been used.
problems become prevalent during sport and cultural events, Before the solution itself, an in-depth analysis of the cur-
as well as in the proximity of administrative buildings or rently used solutions and approaches has been performed.
banks. Parking in the towns and cities during the traffic peak The findings of this analysis are provided below.
times also poses a significant problem. A lack of knowledge
of current number of parking spaces can lead the drivers to 2. Related Works
fully occupied parking lots, which consequently leads to
having to move to another location, along with searching for Firstly, an analysis of the approach to the smart parking
a spot on another parking lot. Therefore, not only do the solution must be performed. The article IoT-Based Smart
drivers waste their time, but also fuel, and it all causes the Parking System [1] is a prominent input in this area, which is
deterioration of the traffic situation and negatively affects the similarly to [2] and describes the architecture of a smart
environment. parking system based on the IoT (Internet of Things)
The aim of this article is to propose a complex smart technology. In [1], sensors placed on the parking spaces
parking lot system based on the mini PC platform, along detect the proximity of the vehicle and send these data to
2 Mobile Information Systems

cloud using mini PC (Raspberry Pi) deployed on the parking in [7] and the spread of the networks’ usage is discussed in
lot. The authors describe the use of Raspberry Pi GPIO pins, [8, 9].
to which 26 sensors can be connected. The number can be
increased even further using a suitable multiplexor. This 3. The Proposed System Architecture
solution utilizes an IBM MQTT server, to which the
Raspberry sends data from the sensors via MQTT messages. The aim of the proposed parking system is a complex and
The user communicates with the system via a mobile app economic system for detection of free parking spaces on an
written in Apache Cordova, which communicates with the outdoor parking lot, which is based on the mini PC
server with messages in the JSON format. The user can use platform. The system partially handles the problems with
the mobile app to see the number of the free spots on the undisciplined drivers who would use the parking lot
parking lot, make a reservation of a parking space, and pay without the respective authorization. This is provided by
the parking fee. In [2], besides the general architecture, the using the automatic gate system at the parking lot entrance.
authors focus on using Elliptic Curve Cryptography (ECC) Because of the budget requirements, the variant where the
as an attractive alternative to the conventional public key parked vehicles would be detected via ultrasound or infra-
cryptography such as RSA. Interesting algorithms employed sensors has been abandoned, and therefore, the proposed
in the contemporary smart parking systems are presented in solution discusses the variant using magnetometric sensors
[3]. Finally, an overall overview of the approaches and so- connected to a mesh network. In regular intervals, the mini
lutions is provided by [4]. PC checks whether the individual parking spaces are oc-
Another topic covered by this article is the optimization cupied and stores the data in a real-time database. The
of the logic and IoT approaches used to find and navigate to driver is then informed via the mobile app for Android, and
the parking space itself. Tsai et al.[5] describe the use of a so they have current information about the numbers of
mobile app and Internet of Things (IoT) technology in the available parking spaces on the selected parking lot. The
parking system: the process of finding a parking lot, parking availability can be checked either manually or automati-
space reservation, and indoor navigation inside the parking cally using the Geofencing service. The user is also able to
lot. Another topic is finding the suitable algorithm to cal- book a parking space for 60 minutes. If they do not arrive at
culate the priority for recommending a specific parking lot the parking lot during this time, the reservation is auto-
based on the distance of the driver from the parking lot, matically canceled.
number of free parking spaces, and the parking fee, with the The proposed solution enables opening the parking lot
preferences being set up by the user. After selecting the gate directly from the mobile app by adding an entry into the
parking lot, the driver has an opportunity to book the real-time database. After the entry is added, the mini PC
parking space via the app. For the calculation of the distance performs the corresponding steps. To check the presence of
from the parking lot and the navigation, the app uses GPS the driver (their smartphone, to be precise) on the parking
coordinates acquired from the Google API. Detection of the lot, the mobile app compares the GPS coordinates before
presence of the vehicle on the parking space is realized via an registering the vehicle as present on the parking lot. If the
ultrasound sensor and transferred to the server via a WiFi distance between the smartphone and the parking lot is
module. The described solution is designated mainly for higher than a given limit, the access is denied. Thanks to this
indoor parking lots, with the indoor navigation provided by solution, there is no necessity for use of an RFID chip, and to
iBeacon Bluetooth transmitters. The presence of the driver make use of the parking lot, a smartphone with the installed
on the parking lot is detected by an RFID chip. The matter of app and connection to the Internet should suffice. The
parking specifically on the side of the roads is discussed in identification of the smartphone is performed using a unique
[6]. It describes the detection of the roadside parking spaces 64-bit number, ANDROID ID. The general model of the
using sensors placed on the vehicle on the passenger’s side. proposed solution is depicted in Figure 1.
These sensors are placed on public transport vehicles, taxis,
or sometimes on the vehicles of the volunteers who fre-
quently pass through the marked measured areas. 3.1. Functional Diagram of the Proposed Solution. The mobile
The system uses an ultrasound detector to detect the app monitors the entry in the database with the number of
parked vehicles and empty spots along the road in con- empty spots. On every change, the state of the notification
junction with the GPS system, and by using Map Matching icon on the app’s main screen is changed. After tapping this
(which compares the detected spots with the map), a map icon, the current occupancy of the individual parking spaces
of available parking spaces is continuously generated and is loaded from the database, and it is displayed on the
is distributed to the users via the mobile app or a website. smartphone screen. By tapping on the respective icon on the
According to the performed measuring, to map the same main screen, navigation to the parking lot is started. Using
number of the parking spaces, mobile sensors are more Google API, the application starts navigation on Google
efficient than sensors placed directly on the parking Maps. More detailed information is described in Section 5.
spaces. The success rate of the detection was between 76 To enable the automatic notifications about parking lots in
and 94 percent, depending on the accuracy of the GPS proximity and the number of available parking spaces, the
system. application uses Geofencing service, which is described in
In the hereby presented complex solution, IQRF net- Section 5.2.
works are utilized. Their communication system is discussed The following diagrams describe two main functions:
Mobile Information Systems 3

IQRF

IQRF

Figure 1: General system structure design.

Displaying the available spots at the parking lot and 4. IQRF Communication Platform
their current occupancy (Figure 2)
Turning on the automatic notification about available From the principle of the intelligent parking system pro-
parking lot (Figure 3) posal, it is necessary for the system itself to be able to detect
whether the respective physical spot is occupied or not, and
The following diagram (Figure 4) describes the logical it must also be connected to the suitable network so it can
steps required for opening the parking lot gate. First, the communicate with the central mini PC. For this purpose,
application checks the database of the parked vehicles to IQRF technology has been chosen [10].
determine whether the vehicle is entering or leaving the IQRF is a platform suitable for wireless data transfer,
parking lot. which utilizes wireless data transfer, using the frequencies of
When entering the parking lot, it is also checked whether 868 MHz and 433 MHz for the communication. The IQRF
there are available spots or if the driver who is sending a platform makes use of special transceivers that reciprocally
request to open the gate has a reservation saved in the interchange the data. The IQRF transceivers are known to
database. If everything is in order, the distance from the have a very low consumption rate (12.3 mA during com-
parking lot is checked to prevent an unintentional or de- munication and 380 nA in sleep mode), and thanks to the
liberate gate closure at higher distance from the parking lot. supported MESH topology, communication at relatively
If any condition is not met, the driver receives information long distance is possible. These parameters appear to be an
about the parking lot unavailability. Otherwise, the gate ideal solution for the IoT technologies. The wireless IQRF
opens, and the arriving vehicle is added to the parking lot’s network uses the IQMEST protocol [11, 12], which uses the
database. At the same time, the driver’s reservation validity is principle that in any given area, and there are always at least
checked and deleted from the database if invalid. When two IQRF transceivers within the transmission reach. The
leaving the parking lot, the system opens the exit gate and maximum distance between two communicating trans-
deletes the record of the vehicle from the parking lot’s ceivers is around 500 meters in space without obstacles. The
database. transceivers transmit in synchronization, so they do not
The realization of the reservation process is depicted in interrupt each other during the transmission. The com-
Figure 5. During an attempt at making a reservation, the munication is governed by a coordinator, which sends the
system first checks whether the vehicle is not currently data, transceivers in the reach receive the data, during their
located at the parking lot and that its ID does not have a time-slot send the data further, and this way the data
reservation created already. Then, the available spots are gradually spread through the whole network. Thanks to this
checked. If all the conditions are met, a reservation entry is principle, the whole network is very reliable and has a high
made in the parking lot’s database. Otherwise, the app success rate of the data transmission. Because of the du-
informs the driver that the reservation cannot be made. plicate paths between individual transceivers, the data reach
Duration of the reservation is limited to 60 minutes. For their destination even if multiple communication paths are
this reason, in 1-minute intervals, the mini PC on the disrupted at the same time. The principle is depicted in
parking lot checks the database if any reservation has Figure 7, where the coordinator C sends the data for the
exceeded this limit. If this occurs, the entry is deleted from transceiver N2. The communication paths that are inter-
the database. rupted, e.g., by signal disruption, are represented by the red
On the general communication schema (Figure 6), you crosses. The data reach their destination, as the green arrows
can see that the common storage for the whole system is a suggest, via the transceivers N1, N3, and N4.
Firebase database. The communication between the database
and the mini PC is procured via TCP/IP protocol. The
communication between the database and the mobile app 4.1. DPA Protocol. Every transceiver has a hardware profile
uses mobile network data transfers and the communication (HWP). Assigning a HWP to the transceivers enables their
between the mini PC and the individual sensors is procured control via messages with DPA (direct peripheral access)
by the IQRF technology described in Chapter 3. protocol [13]. Because of this protocol, it is possible to build
4 Mobile Information Systems

Mobile apps Mobile apps

Command to display the

current occupancy of
parking spaces
Firebase

Read data from

database when
changing value
Firebase
The number of
free parking Reading data
spaces from a
database

Displaying the current

occupancy of parking
spaces

Figure 2: Functional diagram to show the number of available parking spaces and their current occupancy.

Automatic
detection by
geofencing
No
Command to start
automatic parking Launch services Parking in
detection within Geofencing the vicinity
range
Mobile apps
Yes

View notification of
available car park
with free space

Figure 3: Functional diagram to trigger the automatic notification of available car parking.

a network comprised of up to 240 transceivers, and the respective command. Using the DPA handler, it is also
transceivers can be controlled by sending the data in the possible to expand the set of the HW groups described
specific format (Table 1). above, and thus it enables the filtering of control message for
NADR (node address) has the address 0x00 for the the groups of peripheries of the respective type.
coordinator and 0x01–0xEF for the other transceivers and
PNUM (peripheral number) 0x03 EEPROM, 0x08 SPI, 0x0C
4.3. FRC. Fast response command (FRC) is a special co-
UART, etc. PCMD (peripheral command) is designated only
ordinator DPA periphery, which enables sending a com-
by the type of the used periphery; HWPID (hardware profile
mand that can be processed by all the transceivers in the
ID) uniquely determines the functionality of the periphery
network. The moment the transceiver processes the com-
device. If 0xFFFF number is used, the command is executed
mand, it stores the response on the specific position in the
on any HW profile. PDATA (peripheral data) is an optional
message and it passes through the whole network along with
56-byte field for additional command parameters.
the data and collects the responses of individual transceivers.
If we need to get the same information from all the
4.2. Custom DPA Handler. For creating the own logic of the transceivers, which is, in our case, the input from a detector
transceiver, custom DPA handler is used, i.e., it uses code about the presence of a vehicle on a parking space, the use of
written in C, which can be used to define custom user FRC is very practical as it is not necessary to send indi-
periphery and set up its behavior during received a DPA vidually to every transceiver, but sending one command is
command with this periphery’s ID and the ID of the enough. This positively affects not only the amount of the
Mobile Information Systems 5

Mobile apps

Command to open the

barrier

Firebase
Read data from
the database
Car in the Opening the
Yes
parking lot? barrier
Deleting a
vehicle record
No from the parking
database
Firebase
Free places?
Reservation?

Yes
No Finding the distance
from the GPS car
park

Information about Exceeding the

the impossibility of Yes specified
opening the barrier limit?

NO

Opening the barrier

View the current

occupancy of the car
There is a park Entry of vehicle
reservation?
entry
information
Delete Yes
reservation Firebase

Firebase

Figure 4: Functional diagram of the entry and exit gate control.

Information about
the inability to
Y make a reservation N
e o
s
Read from
database Car in the
Booking parking lot?
order Is there a No Free place?
Firebase
reservation?
Mobile apps
Write a
reservation to
the database Yes

Firebase

Figure 5: Functional diagram for making a reservation.

6 Mobile Information Systems

GPRS, EDGE,
LTE ..

TCP/IP IQRF
Firebase
Mobile apps Mini PC Sensor set

Figure 6: Functional diagram to communication.

N5 N2
After programming all the transceivers and inserting
N3 them into the DK-EVAL-04A testing modules or alterna-
tively by connecting custom modules that would need to be
C adjusted according to the transceiver connection schema
(Figure 9), for the transceiver to be powered by the right
N1 N4 voltage and have accessible RESET and bonding buttons, by
connecting the coordinator into the CK–USB–04K pro-
Figure 7: Principle of IQRF communication between coordinator grammer, an IQRF network can be created.
C and N2 transceiver. If a red LED is blinking after connecting the power source, it
means that no previous bonding is stored in the memory.
Otherwise, it is necessary to manually unbond by pressing the
transferred data but also time needed to get a response. The reset and the user buttons on the testing module and then
FRC command is sent via the DPA protocol (Table 2). releasing the reset button. After a green LED blinks, the user
User data item is not required by all the FRC commands, button must be released at once. For erasing the data about
and if not used, it must be replaced by a 2-byte value of 00.00. bonding from the coordinator, IQMESH Network Manager,
SEND command can be replaced by the SEND selective which is a part of the IQRF IDE program, must be used. To do
(0x02) command, which makes it possible to send the so, press the “Clear All Bonds” button. Afterwards, by pressing
command only to the chosen transceivers, which are defined the “Bond” button, the coordinator start searching for a new
in the data field between ID and user data entries. node, and in the frame of ten seconds, the bonding must be
The selected transceivers can be specified through 30- confirmed by pressing the user button on the respective testing
byte binary information. In IQRF network, there can be the module. By repeating this procedure, all the nodes must be
maximum of 240 transceivers, which corresponds to 30 bonded with the coordinator. At the time of the bonding, all the
bytes, i.e., 240 bits. If the respective bit has a value of 1, the nodes must be in the communication reach of the coordinator.
command will be sent to the corresponding transceiver. After the bonding is finished and the respective nodes are placed
Otherwise (the value of 0), the transceiver will not be in- on their final positions, by clicking the “Discovery” button in
cluded into processing of the command. IQMESH Network Manager, the IQMESH network topology is
For the initial configuration of the transceivers and for created. It can be viewed in the “MapView” tab (Figure 10).
the creation of the IQRF network, it is necessary to use
CK–USB–04K programmer and IQRF IDE development
interface. It is also necessary to get hardware profiles for the 4.4. IQRF Gateway. The created IQRF network requires a
coordinator and the individual nodes. gateway for transmitting the data to the database/cloud via
Into the newly created project, the author must add the the Internet. For the purposes of the best compatibility and
HW profiles and the DPA custom handler for FRC com- technical support, a one-board computer UpBoard has been
mand for detecting the state of the sensors on the parking used as it, in comparison to Raspberry Pi, offers a micro-
spaces. The HW profiles belong to plug-ins section and the processor with better performance in Intel Atom, as well as
DPA custom handler to source section. Considering the higher memory capacity, and does not require an OS to be
fact that the inserted DPA handler is in the format of the installed on an external memory card, which can pose po-
source file written in C, this file must be compiled to ∗ hex tential problems in conjunction with the mechanical con-
format so it can be uploaded to the transceiver. After ® ™
nector. UpBoard is also used in Intel RealSense Robotic
Development Kit and has the following characteristics:
inserting the first transceiver into the programmer and
connecting it to a USB port of a PC, in the “Project”
window in the “TR Configuration” section, the configu- ® ™
Intel Atom x5-Z8350 SoC
Onboard DDR3L Memory up to 4 GB
ration settings for the inserted transceiver can be opened.
For all the communicating transceivers, it is necessary to set Onboard eMMC Storage up to 64 GB
the same communication channel (typically 52) in the “OS” Gigabit LAN × 1, USB 2.0 × 4, USB 3.0 × 1, HDMI × 1
tab. Next, in the “HWP” section, the option to process the 5V DC-in
FRC commands for the coordinator and use of custom
40 pin GPIO × 1
DPA handler for every node must be enabled. In the
“Security” tab, a password and communication encryption DSI/eDP × 1
can be set up (Figure 8). MIPI-CSI × 1
Mobile Information Systems 7

Table 1: DPA packet structure.

PNUM (peripheral
NADR (node address) PCMD (peripheral command) HWPID (hardware profile ID) PDATA (peripheral data)
number)
[2B] [1B] [1B] [2B] [0–56B]

Table 2: FRC (SEND) command structure.

PCMD (peripheral HWPID (hardware
NADR (node address) PNUM (peripheral number) PDATA (peripheral data)
command) profile ID)
0x00 0x0D 0x00 0xFFFF ID User data
Coordinator FRC SEND All groups Command ID Data for FRC processing

for programming of IQRF network jobs, a database

management, a NodeRED, was utilized.

4.5. Android Geofencing. The last component used in the

presented solution is the Android Geofencing service, which
serves to alert the driver about the availability of a parking lot
within the mobile application. This service uses the capa-
bilities of the mobile phone to determine its current location
using GPS, the availability of known WIFI networks, and
the distance from the mobile operator’s BTS based on the
signal strength of the mobile network. To use the service,
firstly, it is needed to enter the latitude and longitude of a
specific place (parking lot) and the radius of the circle
centered in this coordinate. The circle created in this way is
called geofence, and then it serves for detecting notifica-
tions or other actions such as turning on Bluetooth and
turning off the phone’s ringtone. The notification can be set
for a situation when the phone enters a defined circle (we
Figure 8: Example of HWP setting for “NOD” transceiver. use this notification in the presented solution), or if it leaves
the circle or stays in it for a certain period of time. Each
Android user can have up to 100 geofences (circles) reg-
istered in all applications on their phone. If the phone is
Considering the availability of the technical support,
located at the intersection of several geofences, it can
Ubilinux OS has been chosen for the smart parking system.
perform actions separately for each of them, or a notifi-
The chosen mini PC serves for controlling the IQRF net-
cation of the availability of multiple geofences (parking
work and for conjoining the whole app via the Internet and
lots) and the distance to the center of each can be sent and
the database. For those reasons, the IQRF coordinator had
calculated from the difference of two GPS coordinates
to be connected using an available reduction with
(phone and geofence).
UpBoard’s GPIO pins, and corresponding utilities for the
communication with the IQRF network and the database
had to be installed as well. For interchanging the infor- 5. Implementation of Key System Functions
mation between the UpBoard and the IQRF network, an In Section 5.1, the solutions described above, including the
MQTT Broker has been used. For the administration, we used IQRF network and NodeRED and the main component
have chosen a web app for IQRF Daemon and NodeRED of the Android app are documented in detail.
environment that will serve for programming the IQRF
network maintenance. For the IQRF gateway we have used
the following software setup. As an MQTT broker, we have 5.1. Vehicle Detection. Vehicle detection is realized via
decided to use the mosquito and mosquito-clients pack- MPU-9250 magnetometric sensor, which is connected via
ages. Furthermore, the dirmngr server was used for cer- I2C bus with an Arduino Mini microprocessor unit
tificate management. The IQRF Gateway Daemon package (Figure 11).
was used as an open-source IQRF gateway solution which is For the communication with the sensor, MPU9250_a-
being widely used in this area, and it is supported also by sukiaaa.h has been used as it enables a simple maintenance of
Raspberry, Belagone, traditional PC and others. Node.js the magnetometric sensor. After the initial establishment,
open-source server platform was used for code execution; the measured values are stored for a reference and then
8 Mobile Information Systems

10
6

8
MCP73831MC Ref.
USB 5
1 VIN 1 3 VBAT 3.3V 4
VIN VBAT LDO C1 C5
2 2 4 JP1 3
F1 C2 C6
SW2
3 SN035-16 2
6 C3 C7
LED3 GND +
4 5 9 1
Accu LED2 C4 C8
5 8 3.6V 9
SIM 7
XC3 IC1
XC1
SW1

LED1

Figure 9: DK-EVAL-04A test module connection.

MPU-9250 I2C Arduino IQRF tranceiver

Digital out Digital in

Figure 11: Block diagram of vehicle detector.

vehicle detection and its subsequent departure from the

parking space, new rest values are saved as referential in
order to reduce the amount of unprompted or erroneous
detections caused by long-term changes of the magnetic field
intensity. The implementation of the vehicle detection
Figure 10: Example of network topology in IQMESH Network
through the change of the magnetic field is described below.
Manager.
void testSensor(){
compared with the currently measures values in the 500- deltaMx � abs(normalMx-mX);
millisecond intervals (during the initialization, the parking deltaMy � abs(normalMy-mY);
space must be empty). The measured values are transferred deltaMz � abs(normalMz-mZ);
to the serial port for the purpose of checking the func-
Serial.println(“deltaMx: ” + String(deltaMx));
tionality of the sensor and checking the measured values. See
the following code for the main program enabling the Serial.println(“deltaMy: ” + String(deltaMy));
communication with the magnetometric sensor: Serial.println(“deltaMz: ” + String(deltaMz));
void loop() { if(deltaMx > 7 || deltaMy > 7 || deltaMz > 7){
if(start){delay(5000);} Serial.println(“Car is present”);
mySensor.magUpdate(); digitalWrite(13, HIGH);
mX � mySensor.magX(); detect � true;
mY � mySensor.magY(); }
mZ � mySensor.magZ(); else
if(start){initSensor(); start � false;} {Serial.println(“Car is not present”);
Serial.println(“magX: ” + String(mX)); digitalWrite(13, LOW);
Serial.println(“maxY: ” + String(mY)); if(detect){start � true; detect � false;}
Serial.println(“magZ: ” + String(mZ)); }
testSensor(); }
Serial.println(“”); // Add an empty line The digital output of the Arduino module is connected to
delay(500); the IQRF transceiver, which provides the connection of the
individual sensors into the IQMESH network.
}
If a metal object (a vehicle) is detected in the proximity
and the values in any of the axes exceed the given limit, the 5.2. Car Parking with NodeRED. As it has been already
digital output signalizes the presence of a vehicle. After every mentioned, the database communicates not only on the
Mobile Information Systems 9

mobile app but also the mini PC via NodeRED. For greater for the gate to open is stored in reservationID item. This ID is
clarity, individual program flows are split into individual then compared to the records in reservations, and if a valid
IQRF_Request tabs. The commands are sent to the IQRF reservation is found, it is deleted. After the time delay, the
network, IQRF_Response processes the responses from the reservationID item is reset by ResetReservID.
IQRF network, Firebase communicates with the Firebase The second flow starts with the GateID node, which
database, Gate operates the entrance and exit gate, and continuously checks the database for changes in the gateID
Reservations maintains the user reservations. item. Into this item, the mobile app saves the ID of the
smartphone registered on the parking lot (it has a record in
carInPark) and sends a request to open the exit gate. The
5.3. IQRF_Request. The program flow IQRF_Request program continues with the link to IQRF_Request, where it
(Figure 12) uses program nodes on the lies 2 through 5, opens the exit gate via the GateExitOpen node. After a time
which send the requests to open or close the entrance or exit delay for the vehicle to pass through, the gate is closed again
gate. The requests come from other tabs via the connections via the link to GateExitClose. The flow continues with
(grey arrows at the beginning of the lines). loading the CarInPark items from the database and coverts
Individual requests are realized via custom functions them to the JSON format. These data are passed to the
written in JavaScript. That is the DPA command for the FirebaseConvert function (code: finding and passing the
IQRF network (code: create a request for an IQRF network). item ID for deletion), which, from the smartphone ID stored
Because of the physical absence of a gate, for the tuning in GlobalContext, searches for a record in the database and
purposes of the app, the commands were simulated by turn deletes the record via the DeleteValue node. The last step of
on or off a red LED on the IQRF coordinator (entry gate) the program is the gateID item reset.
and a green LED (exit gate). In practice, the LED would be
replaced by a digital output that would send the command to
the gate. 5.5. Reservations. In the case of this flow, a division into two
var data � { parts has been done too (Figure 14). Every minute, the first
part checks the reservation length, and in case of exceeding
type: “raw”,
the time limit, the reservation is removed. The other part
request: { checks the reservations of the drivers entering the parking
nadr: “0x0000”, lot, and if such an entry exists, it is deleted.
pnum: “0x06”, Every minute, the inject node loads the reservations from
the database and converts them to JSON. The data are passed
pcmd: “0x01”,
by the FirebaseConvert (code: check out booking timeout)
hwpid: “0xFFFF”, function, which checks the length of every reservation and in
pdata: “”, case of exceeding the time limit and passes the respective
}, reservation for deletion. If a reservation that should be
deleted is found, the switch continues with the second path.
timeout: 1000 Otherwise, it continues with the first path, where the pro-
} gram ends.
msg.payload � data; var response � msg.payload;
return msg; var data � “”;
var nic � 1;
5.4. Gate. For the communication itself and its procure- var data1 � response.split(“{”).toString();
ment, the gate is a key component of the whole architecture.
The gate process flow is split into two parts. The first part var l � data1.length;
controls the entry gate, and the other controls the exit gate for(i�0;i<l;i++){
(Figure 13). var s � data1.substring(i, i + 1);
The first flow begins with the Gate node, which is, in fact,
if(s �� “\””){}
a listener that watches the Gate item in the database. It
follows with the switch on the value of 1 (opens the gate) and else if(s �� “{”){}
continues with the second path to the GateEntranceOpen else if(s �� “}”){}
node. The next node is a time delay for the vehicle to pass else {data � data + s;}
(sensors for the detection of a vehicle in the gate space are its
}
part) and then the flow continues with the link to Gate-
EntranceClose and sets the value of the gate item in the var allmsg � data.split(“,−”);
database to 0. In that case, the switch ensures that the gate l � allmsg.length;
does not open again after the value in the database is for(i�1;i<l;i++){
changed when the value of 0 is added, and so it takes the path
var data1 � allmsg[i].split(“:”);
1, to which no other program node is connected. After
opening the entry gate, the ID of the smartphone that asked var fh � parseInt(data1[4]);
10 Mobile Information Systems

Inject Request JSON Gateway request

connected
Gateentranceopen

Gateentranceclose

Gateexitopen

Gateexitclose

Figure 12: Program flow IQRF_Request.

Gate Switch Delay 2s ResetreservID

Ready Ready

Delay 10s Gate

Ready

Delay 10s

GateID CarInPark Json FirebaseConvert

Ready Ready

DeleteValue
GlobalContext Ready

GateIDUpdate
Ready

Figure 13: Program flow gate.

1 Reservations Json
Ready

Firebaseconvert Switch Deletevalue

Ready

Globalcontext Reservations Json Reservationsconvert

Ready

ReservationID Switch Deletevalue

Ready Ready

Figure 14: Parking reservation management.

var fm � parseInt(data1[5]); var cesta � “parkings/parking1/reservations/

var date � new Date(); −” + data1[0];
var h � date.getHours()+1; node.send({childpath:cesta});
var m � date.getMinutes(); }
var delta � (h ∗ 60 + m)−(fh∗60+fm); else{
if(delta > 60){ node.send({payload:nic});
Mobile Information Systems 11

} parking lot is procured via geofencing service For this

} purpose, the Constants class is used. It contains the pa-
rameters such as creating a geofence (GPS coordinates and
return;
radius). Using a hash map, several geofences can be created
As mentioned previously, when the entrance barrier is simply by adding a name and coordinates via another
opened, the phone ID is written to the reservationID entry. The command.
change of its state is monitored by the ResourceID node, which
LANDMARKS.put(. . .. . .)); (code: set geofence
begins the second part of this flow. The phone ID is stored in a
parameters).
global variable, and all reservations are read from the database,
converted to JSON format and passed to the Reser- public class Constants {
vationConvert (code: delete driver’s reservation for parking) public static final float GEOFENCE_RADIUS_
function, which compares the records in the database with the IN_METERS � 1000;
stored phone ID. When a match is found, the entry is deleted. public static final HashMap < String,
The switch performs the same function as in the previous case. LatLng > LANDMARKS � new HashMap < String,
var response � msg.payload; LatLng > ();
var reservationID � global.get(“reservationID”); static {
var data � “”; // Parking 1
var data1 � response.split(“{”).toString(); LANDMARKS.put(“Parking 1”, new LatLng
(50.420860, 16.185796));
var l � data1.length;
}
for(i�0;i<l;i++){
}
var s � data1.substring(i, i + 1);
if(s �� “\””){} By calling the populateGeofenceList() method in
MainActivity, custom geofence is created (code: creating
else if(s �� “{”){}
geofence).
else if(s �� “{”){}
public void populateGeofenceList() {
else {data � data + s;}
for (Map.Entry < String, LatLng > entry: Con-
}
stants.LANDMARKS.entrySet()) {
var allmsg � data.split(“,−”);
mGeofenceList.add(new Geofence.Builder()
var al � allmsg.length;
setRequestId(entry.getKey())
for(i�1;i<al;i++){
setCircularRegion(
var data3 � allmsg[i].split(“,”);
entry.getValue().latitude,
var androidID � data3[1].substring(10, data3
entry.getValue().longitude,
[1].length);
Constants.GEOFENCE_RADIUS_IN_
var path � “parkings/parking1/reservations/” +
METERS
data3[2].substring(3, data3[2].length);
)
if(androidID �� reservationID){
setExpirationDuration(Geofence.
node.send({childpath:path}); NEVER_EXPIRE)
else {node.send({payload:1});} setTransitionTypes(Geofence.GEOFENCE_
} TRANSITION_ENTER)
return; build());
}
5.6. Smartphone Application. For the purposes of the testing }
and validating, an Android app has been developed. This app The GeofenceTransitionsIntentService class subse-
enables control of all the functions on the screen, so the quently creates a notification channel that it uses to inform
controls are user-friendly and simple. The functions are the driver with a notification about available parking lot
started by tapping simple icons, so the driver is not forced to upon entering inside the created geofence (Figure 15).
read through the context menus while driving. The overview
of all the functions and meaning of the individual icons is
represented by the image. 5.8. Navigation to the Parking Lot. To run the navigation, the
app uses Google API and Google Maps for Android, which
enables running a map in the following modes:
5.7. Automatic Detection of a Nearby Parking Lot. As de-
scribed in chapter 4.5, the automatic detection of a nearby Display the map on a given place with given zoom level
12 Mobile Information Systems

@Override
public void onDataChange(@NonNull DataSnap-
shot dataSnapshot) {
places � dataSnapshot.getValue().toString();
int pl � Integer.parseInt(places);
int cr � (int) (pl − countRes);
Fplaces.setText(String.valueOf(cr));
}
@Override
public void onCancelled(@NonNull DatabaseError
Figure 15: Notification after entering geofence.
databaseError) {
places � “Error”;
}
Searching for a place and displaying it on the map });
Navigation with the selected means of transport (car,
bicycle, and walking) On every change of the available parking spaces, the method
above (code: setting the listener to change the number of free
Displaying the panorama view in Google Street View parking spaces) sets the textView value on the main activity via
service the setText method, so the button always contains the current
To run the map, it is first necessary to create an “Intent” number of available spaces on the parking lot. When tapping
object and specify the mode in which the map will be this button, a new activity is run. It contains listView, which
opened. Intent contains a special string (URI), which ac- informs the driver about current occupancy of the individual
curately specifies the requested action. After creating the parking spaces on the parking lot. The same activity is run also
Intent, the activity gets started by the startActivity() method. upon entering the parking lot after the request to open the entry
As you can see in the following code (code: create Intent and gate, so the driver has an overview of the available spaces. For
run Map Activity), Intent is created via URI, which defines this purpose, the app uses, similarly to the previous case, the data
the type of the action as navigation to a set GPS coordinate. stored in the database, whose changes are monitored by the
Subsequently, the Intent is injected with a packet to secure listener. The data about the occupancy of the individual spaces
that is processed by Google Maps. are stored in the database via NodeRED in the form of a string
“0000110100110010”, which contains sixteen bits: “1” or “0”. If
public void startNavigationButtonHandler(View view) the given position has the value of “1”, it means that the space is
{ occupied. Otherwise, the space is available. The getOccupancy()
Uri gmmIntentUri � Uri.parse(“google.navigation: method (code: filling a text field based on database data) in this
q � 50.475367, 16.179489”); activity fills the OCCUPANCY[] array with the values “FREE”
Intent mapIntent � new Intent(Intent.ACTION_- or “OCCUPIED”.
VIEW, gmmIntentUri); public void getOccupancy(){
mapIntent.setPack- for(int i�0;i < 8;i++){
age(“com.google.android.apps.maps”) ;
System.out.println(places.substring(i, i + 1));
startActivity(mapIntent);
if(places.substring(i, i + 1).equals(“0”)){OCCU-
}
PANCY[i] � “FREE”;}
if(places.substring(i, i + 1).equals(“1”)){OCCU-
5.9. Information about the Parking Lot Occupancy. The in- PANCY[i] � “OCCUPIED”;}
formation about number of available and occupied parking if(places.substring(i + 8, i + 9).equals(“0”)){OC-
space is saved by NodeRED into the Firebase database, and in CUPANCY[i + 8] � “FREE”;}
the Android app, the instance of the database is created: if(places.substring(i + 8, i + 9).equals(“1”)){OC-
FirebaseDatabase database � FirebaseDatabase.getInstance(), CUPANCY[i + 8] � “OCCUPIED”;}
along with the references for its individual items: Data-
baseReference myRef. . . � database.getReference(“parkings/ NUMBERS[i] � i + 1;
parking1/. . .. . .. . .”). This is followed by the listener per- }
forming the corresponding actions every time a value changes }
in any of the values it is monitoring.
Afterwards, these values are in the adapter using listView
public void readFreePlaces(){ passed to individual textView (code: filling text boxes and
myRefFplaces.addValueEventListener(new Val- setting font color based on content), as is shown in the image
ueEventListener() { (Figure 16).
Mobile Information Systems 13

Figure 16: Parking occupancy is displayed.

public View getView(int i, View view, ViewGroup androidID–unique 64-bit number generated by the
viewGroup) { Android OS
view- time–the time when the reservation is created
� getLayoutInflater().inflate(R.layout.customlayout,
Before creating a reservation, based on the Android ID, the
null);
app checks whether the user is already saved in the database
TextView (has a valid reservation or is already on the parking lot) (Sample
tvNumbers � view.findViewById(R.id.tvNumbers); 14) and if there are any spaces available for reservation on the
TextView parking lot (number of free spaces—number of valid reser-
tvOccupancy � view.findViewById(R.id.tvOccupancy); vations). If a reservation is possible, it asks the database for a
TextView tvOccu- new item id: id � myRefRes.push().getKey(). It then creates a
pancy1 � view.findViewById(R.id.tvOccupancy1) ; new instance of the Reservation class: myRe-
fRes.child(id).setValue(reservations; and finally, this newly
tvNumbers.setText(String.valueOf(NUMBERS
created instance is stored in the database (Figure 17): myRe-
[i]));
fRes.child(id).setValue(reservations) (code: determining the
tvOccupancy.setText(String.valueOf(OCCU- status of a user’s reservation).
PANCY[i]));
public void onDataChange(DataSnapshot snapshot) {
tvOccupancy1.setText(String.valueOf(OCCU-
PANCY[i+8])); countRes � snapshot.getChildrenCount();
if(OCCUPANCY[i].equals(“FREE”)){tvOccu- updateCountReservation();
pancy.setTextColor(Color.parseColor(“#00ff00”));} resIndicator � false;
if(OCCUPANCY[i].equals(“OCCUPIED”)){tvOccu- for (DataSnapshot postSnapshot: snapshot.getCh-
pancy.setTextColor(Color.parseColor(“#ff0000”));} ildren()) {
if(OCCUPANCY[i+8].equals(“FREE”)){tvOccu- Reservation
pancy1.setTextColor(Color.parseColor(“#00ff00”));} post � postSnapshot.getValue(Reservation.class);
if(OCCUPANCY[i+8].equals(“OCCUPIED”)){tvOc- String resCheck � post.getAndroidID();
cupancy1.setTextColor(Color.parseColor(“#ff0000”));} if(resCheck.equals(androidID)){resIn-
return view; dicator � true;}
} }
} }
The state of the reservations is regularly monitored also
by the mini PC via NodeRED, and if the time limit for the
5.10. Creating a Parking Space Reservation. Every driver can
reservation is exceeded, it is automatically removed from the
make one short-term reservation of one parking space. For
database. The reservation is removed also if a driver with a
this purpose, the app uses the Reservation.java class, which
valid reservation arrives on the parking lot and sends a
has three attributes:
request to open the entry gate. The procedure for removing
id–item ID generated by the Firebase database the reservation from the database is described later.
14 Mobile Information Systems

Parkings
Reservations
Parking1
-LYY5_Y2ueZ33fHK2Zq8
CarInPark

AndroidID: “fd5ee236f37652ff -LYhBh5zxFsxum8tn2Y4

AndroidID: “fd528236f37652aa
Id: “-LYY5_Y2ueZ33fHK2Zq8
Id: “-LYhBh5zxFsxum8tn2Y4
Time: “20 : 06” Time: “19 : 09”
Figure 17: Example of saving the reservation to the database. Figure 18: Record the vehicle present in the parking lot.

5.11. Opening the Entry/Exit Gate. To open the entry/exit

gate, the app again uses the Firebase database, which con- Arduino module program was modified to count the vehicle
tains two items: detections. The testing was carried out for the period of 24
gate: it can contain two values: 1 and 0 (entry gate open/ hours, performing a check twice per second. The sensor was
closed) placed and fixed so its move in any direction was impossible.
In 24 hours, 172,800 checks were performed, and only four
gateID: in case the car is present on the parking lot, the
detections were faulty, with the error rate of 2.3%.
app saves the androidID of the user who wants to open
In the second phase, correct detection of five vehicles of
the exit gate
different sizes (Toyota Yaris, Škoda Roomster, Škoda Octavia
The app contains the CarInPark.java class, which has, combi, Citroën Jumpy, and Nissan Navara) was tested. Every
similarly to the Reservation.java class, three attributes: vehicle was placed thirty times over the magnetometric
sensor. The detection success rate was very high—there has
id–item ID generated by the Firebase database
been only one faulty detection, which happened with the
androidID–unique 64-bit number generated by the smallest of the vehicles (Toyota Yaris).
Android OS The automatic nearby parking lot detection has been
time–time of arrival on the parking lot carried out over the course of two weeks. The first week,
testing for choosing the appropriate radius of the geofence
The same as with the requirement to create a reservation,
was carried out, so the driver would get an information
during the requirement to open the gate, the app first checks about an available parking lot in time and would be able to
whether the vehicle is currently present at the parking lot respond to this information easily. After one-week testing,
(has a record in the database) via androidID. Then, by 1,000 meters was chosen as the most suitable radius. Because
comparing two GPC coordinates, it checks whether the
of the delay described earlier, the driver in town traffic would
maximum distance of the smartphone from the gate is not
the information about 500 to 800 meters before the parking
exceeded. If these conditions are met, by entering the value
lot. The second week, after setting the suitable radius, testing
of “1” to the gate item of the Firebase database, NodeRED
on a 10-kilometer route was carried out. On this route, five
procures that the entry gate is open. Then, it follows with a
geofences were set up. Ten drivers have driven in this route
request for a new id to save the item in the database: String
ten times in total, which means that 500 tests have been
id � myRefCar.push().getKey(); then, a new instance of the
performed. In three cases, the notification was not received,
CarInPark class is created: carInPark � new CarInPark(id,
and in four cases, the driver received the notification in
androidID, time); and finally, the item is saved in the da- distance under 200 meters from the parking lot. This adds up
tabase: myRefCar.child(id).setValue(carInPark) (Figure 18). to the error rate of 1.4%.
After closing the gate, using NodeRED, a potential reser- One of the important factors for successful adoption of
vation belonging to the driver entering the parking lot is any car parking solution is the economic efficiency. The
deleted.
financial costs of traditional and complex solutions, which
If the vehicle is on the parking lot (user’s androidID is in
have been stated in related works, are certainly higher than
the database), androidID is saved in the gateID item, and
the low-cost solutions based mini PC. The final costs are also
based upon this action, NodeRED opens the exit gate and
affected by the selection of individual components, so the
removes the corresponding user’s record. The processes of
final cost of the solution was not high. The approximate
opening and closing the parking lot gates using NodeRED
retail prices, at which the individual components of the
have been described in Chapter 6.3.
system were purchased, are summarized in the following list:
MPU–9250 (3 €)
6. Discussion
Arduino Mini (2 €)
During the pilot operation, vehicle detection using a mag- IQRF TR-72D (14 €)
netometric sensor has been tested in two phases. In the first
Battery (19 €)
phase, unprompted detection testing has been performed. In
the second phase, testing for detection of vehicles of various UbBoard-mini PC (90 €)
sizes was followed. For the first phase of the testing, an Entrance barrier (606 €)
Mobile Information Systems 15

7. Conclusion [2] I. Chatzigiannakis, A. Vitaletti, and A. Pyrgelis, “A privacy-

preserving smart parking system using an IoT elliptic curve
The presented solution has been used for pilot deployment, based security platform,” Computer Communications,
and in case the solution was to be used commercially, some vol. 89–90, pp. 165–177, 2016.
enhancements of its functionality would be needed. In the [3] S. Kubler, J. Robert, A. Hefnawy, C. Cherifi, A. Bouras, and
current state, since the time is saved in the database upon K. Främling, “IoT-based smart parking system for sporting
event management,” in Proceedings of the 13th International
vehicle entering the parking lot, the app would be capable of
Conference on Mobile and Ubiquitous Systems: Computing,
calculating the parking fee if price tariffs were to be added. Networking and Services-MOBIQUITOUS 2016 [online],
For the commercial use, it would be necessary to implement pp. 104–114, New York, USA, November 2016.
a way to pay the parking fee, which could be done either [4] H. Arasteh et al., “Iot-based smart cities: a survey,” in Pro-
using a parking machine placed directly on the parking lot or ceedings of IEEE 16th International Conference on Environ-
implementing a connection to a payment gateway to the app. ment and Electrical Engineering (EEEIC), pp. 1–6, Florence,
It would also be suitable to also develop an iOS app, so that Italy, June 2016.
the iPhone users could use the system as well. Another [5] MF. Tsai, Y. C. Kiong, and A. Sinn, “Smart service relying on
option would be to develop a web app for communication internet of things technology in parking systems,” The Journal
with the system. The app would run directly in a smartphone of Supercomputing, vol. 74, no. 9, pp. 4315–4338, 2018.
web browser, and therefore, the system would be cross- [6] C. Roman, R. Liao, P. Ball, S. Ou, and M. de Heaver,
“Detecting on-street parking spaces in smart cities: perfor-
platform. The majority of the system parts and the mobile
mance evaluation of fixed and mobile sensing systems,” IEEE
app have been designed in the way that would allow future Transactions on Intelligent Transportation Systems, vol. 19,
development of more functionalities. The structure of the no. 7, pp. 2234–2245, 2018.
data stored in the database makes it possible to add addi- [7] I. Calvo, J. M. Gil-Garcı́a, I. Recio, A. López, and J. Quesada,
tional parking lots, and the automatic parking lot availability “Building IoT applications with raspberry Pi and low power
detection enables to display notifications about more than IQRF communication modules,” Electronics, vol. 5, no. 4,
one parking lot. p. 54, 2016.
Therefore, after some modifications, the system could be [8] V. Jan, P. Martin, and R. Hajovsky, “Wireless measurement of
used by, for example, owners of permanent parking lots, also carbon dioxide by use of IQRF technology,” IFAC-Paper-
during large occasional events. Considering its low expenses, sOnLine, vol. 51, no. 6, pp. 78–83, 2018.
[9] J. Skovranek, P. Martin, and R. Hajovsky, “Use of the IQRF
the system could not only help the drivers solve their
and Node-RED technology for control and visualization in an
problems with searching for empty parking spaces and IQMESH network,” IFAC-PapersOnLine, vol. 51, no. 6,
improve the traffic in the towns and cities, but it could also pp. 295–300, 2018.
help to deal with the problems with the drivers who avoid [10] M. Pies and R. Hajovsky, “Using the IQRF technology for the
paying the parking fees, especially on the open parking lots internet of things: case studies,” in Mobile and Wireless
with the parking machines. Technologies 2017. ICMWT 2017. Lecture Notes in Electrical
Engineering, K. Kim and N. Joukov, Eds., vol. 425, Singapore,
Springer, 2017.
Data Availability [11] J. Kopják and G. Sebestyén, “Comparison of data collecting
methods in wireless mesh sensor networks,” in Proceedings of
The measured data used to support the findings of this study
IEEE 16th World Symposium on Applied Machine Intelligence
are available from the corresponding author upon request. and Informatics (Sami), pp. 000155–000160, Kosice, Slovakia,
February 2018.
Conflicts of Interest [12] O. Vondrouš, Z. Kocur, T. Hégr, and O. Slavı́ček, “Perfor-
mance evaluation of IoT mesh networking technology in ISM
The authors declare that they have no conflicts of interest. frequency band,” in Proceedings of 17th International Con-
ference on Mechatronics-Mechatronika (ME), pp. 1–8, Prague,
Czech Republic, December 2016.
Acknowledgments [13] IQRF Tech sro, IQRF DPA framework Technical Guide,
Vol. 10, IQRF Tech, Jičı́n, Czech Republic, 2018.
This work was supported by a Specific Research Project,
Faculty of Informatics and Management, University of
Hradec Kralove, Czech Republic. We would like to thank
Mrs. H. Svecova, a doctoral student, and Mr. J. Dian, a
graduate, of Faculty of Management and Informatics,
University of Hradec Kralove, for the practical verification of
the proposed solutions and close cooperation in the solution.

References
[1] A. Khanna and R. Anand, “IoT based smart parking system,”
in Proceedings of International Conference on Internet of
Things and Applications (IOTA), pp. 266–270, Pune, India,
January 2016.